Agronomía Colombiana

Description of study area

The research was conducted in both greenhouse and field settings. The greenhouse is located in the residential complex “Los Almendros” at km 4 on the Carepa – Apartadó road, in the municipality of Carepa, Antioquia (Colombia) (Fig. 1). The fieldwork was conducted at the Ramiro Jaramillo Sossa experimental farm located 2 km northwest of the urban center of the municipality of Carepa. Both locations belong to the Banana Research Center (Cenibanano), which is associated with the Colombian Banana Growers Association (Augura). The two experiments (field and greenhouse conditions) were carried out simultaneously and lasted three months from September to December 2022. The agroclimatic conditions of the experimental farm include average annual precipitation of 2961 mm, average temperature 27°C, altitude of 20 m a.s.l., and typical conditions of the tropical rainforest (bh-T) (Instituto de Hidrología Meteorología y Estudios Ambientales, 2022; Jaramillo, 2014).

Figure 1.: Location of the study zone in the Antioquia department (Colombia), WGS 84 Magna Sirgas Colombia West Zone (CRS:3115).

Stage I. Greenhouse conditions

Nine sedimentation boxes were prepared using plastic containers measuring 34, 28, and 20 cm in length (l_cs), width (w_cs) and depth (h_cs), respectively. Each box was filled with 17 kg of soil obtained from the experimental field area, clay loam in texture, classified as Fluvaquentic Eutrudepts. The weed species Peperomia pellucida, Selaginella sp. and Digitaria horizontalis, commonly found in banana crops, were established in a mixture from rooted cuttings in three trays (three replicates). These were watered weekly until complete coverage of the trays was achieved (VCT). In three other trays, the same weed species were established, but conventional pruning was carried out (CMT). The remaining trays were left with bare soil without plants (BST).

A wooden structure was designed to support and arrange each of the trays at a 45° angle to the surface, simulating the average slope of the tertiary channels under field conditions (Fig. 2). They were randomly distributed in the section set up for the greenhouse experiment. A 3 L plastic container was attached to the underside of each box, which acted as a collector. The most frequently occurring precipitation event recorded in the banana region from 2019 to 2021 was 10 mm. This value was used in the experiment, with 10 mm of water applied to each tray weekly for eight consecutive weeks. The excess of water from each tray was recovered and stored (runoff and leachate) to determine the total volume (V_T) for the evaluation period. A subsample of 100 ml (V_s) was then taken and dried in an oven at 60ºC until reaching constant weight. Once this condition was reached, the accumulated soil in the subsample (m_s) was weighed and the total soil removed from the tray during the evaluation period (m_T) was determined using Equation 1:

Figure 2.: Sketch of a tertiary drainage, where y: drainage depth; x: horizontal distance relative to height; h: drainage slope; Cs: sediment collection and delimitation box; lcs: box length; acs: box width; λ: slope length.

Stage II. Field conditions

The drainage system structure in the experimental field consisted of tertiary level drains designed to facilitate water exit from the soil profile; secondary drains removed water from the plots, while primary drains water removed from the farm. In plots 4, 9, and 10, a tertiary drain was selected and three sedimentation boxes of 50 cm by 30 cm long (l_cs) and width (a_cs) respectively were installed in the central part of these plots. The top of the box was left open to allow runoff water collected along the length of slope (λ) to enter (Fig. 2). In each box, one of the treatments described above (VCT, CMT, BST) was randomly installed.

At the beginning of the measurements, a 20 L capacity plastic container was attached to the bottom of each sedimentation box. Each week, percolated water was collected, its volume (V_T) determined, and a 50 ml subsample (V_s) extracted. Using the same methodology as in Stage I, the eroded soil was determined.

Variables

Precipitation. The precipitation data were obtained by a pluviometer located in the experimental field, which had daily records. For the study, data for the period from 2019 to 2021 were analyzed.

Modified photochemical reflectance index (MPRI). An image with a RGB (Red, Green, Blue) camera was captured over the slope of each of the treatments (Pacheco & Montilla, 2021). The sedimentation plots in the image were manually delineated, and the degree of vegetation cover was determined using the MPRI, which relates the green (G) and red (R) bands as shown in Equation 2 (Delgado-Bejarano et al., 2023; Pacheco & Montilla, 2021):

Slope (m_se) and length (λ). The inclination angle of the slope was estimated by measuring the channel depth (h) and the slope length (λ) using tape measures. The slope angle (θ) was obtained by Equation 3:

The length (L_se) of the catchment area along the slope was determined as the distance between the top of the channel and the lower boundary of the sedimentation box.

Soil organic matter content. At the beginning of the experiment, soil sampling was carried out at the point adjacent to each sedimentation box, and the organic matter content of the soil was determined in the laboratory using the method of Walkley and Black (1934).

Soil permeability. The permeability was determined in each sedimentation box in the area adjacent to the soil slope by evaluating the hydraulic conductivity of the soil using Zhang's (1997) method with a mini disc infiltrometer.

Soil texture. This parameter was determined in the laboratory by the hydrometer method (Bouyoucos), finding the values of each soil fraction (clay, silt, and sand) (Jaramillo, 2014). For the greenhouse samples, soil used in the traps was analyzed, while in the field, a soil sample was taken from each trap in the area adjacent to the study channel.

Analysis by the Universal soil loss equation (A)

The Universal soil loss equation (USLE) was used to estimate soil erosion for each treatment (A, t ha^-1 year^-1) (Eq. 4). The USLE incorporates parameters including vegetation cover (C, dimensionless), slope gradient (S, dimensionless), slope length (L, dimensionless), rainfall erosivity (R, MJ mm ha^-1 h^-1 year^-1), soil erodibility (K, t h MJ^-1 mm^-1) and management practices (P, dimensionless):

The vegetation cover factor was calculated using the vegetation index (MPRI). The parameters α and β of the equation were assigned values of 2 and 1, respectively (Eq. 5) (Almagro et al., 2019; Amellah & el Morabiti, 2021; Bai & Cui, 2021; Delgado-Bejarano et al., 2023):

The topographic factor was represented by the length (L) and the slope gradient (S) estimated by using equations 6, 7, and 8, which are based on the parameters m_se and λ, which correspond to the slope exponent and the horizontal projection of the slope, respectively. In equation 6, the parameter m_se takes values of 0.2, 0.3, 0.4, and 0.5 when the slope is less than 1%, between 1 and 3%, between 3 and 4.5%, or greater than 4.5%, respectively. The slope gradient (S) is a function of the slope angle in degrees (θ), which varies depending on whether this value in % (θ) is less than or greater than 9% (Wijesundara et al., 2018; Wischmeier & Smith, 1978).

The erosion caused by rain (R) was calculated as illustrated in equation 9, where P_pi refers to the precipitation of the i-th month (mm), P_j to the annual precipitation of the j-th year (mm), based on the number of years in the time series n_j (Bai & Cui, 2021; Han et al., 2021).

The soil erodibility factor (K) refers to the effect of soil properties on its susceptibility to being eroded or transported due to the impact of raindrops. Equation 10 was used to calculate this parameter (Efthimiou et al., 2020; Mahapatra et al., 2018; Wischmeier & Smith, 1978), considering soil properties of organic matter content (OM), permeability (p), structure (s), and particle size ratio (M) using Equation 11:

The p parameter was determined on a scale of 1 to 6 based on the classification of the United States Department of Agriculture (USDA (1999)). For the variable s, the classification given by Wischmeier and Smith (1978) was used.

Finally, the P factor indicates the management practices that can be implemented to reduce the amount of soil transported or eroded, ranging from 0 to 1, where 0 indicates that the activity completely reduces erosion, and 1 indicates that no conservation practice is done. Wischmeier and Smith (1978) highlighted key conservation practices, such as contour plowing, terracing, and no tillage systems. In the case of drainage networks, no conservation practices are applied, so the factor P is equal to 1.

Statistical analysis

The erosion rates directly measured through sedimentation tanks under both greenhouse and field conditions, as well as those estimated through the EEP, were considered as response variables in each case. These were analyzed based on the treatments (types of coverage) using an analysis of variance test (ANOVA). The normality of the residuals of the model and the homoscedasticity between treatments were evaluated. If the assumptions were not met, the Kruskal-Wallis non-parametric test was applied, followed by a post-hoc LSD test or Tukey test. The correlation between the parameters of the EEP with the erosive potential was determined using a Spearman correlation coefficient.

Greenhouse conditions

The greenhouse evaluations revealed significant differences in soil erosion rates among the different types of vegetation cover (P=0.00533, Fig. 3).

As shown in Figure 3, BST exhibited the highest erosion rate compared to the other treatments, with a mean of 38.16 t ha^-1 year^-1. This value is qualified as intermediate erosion according to the classification given by Han et al. (2021). This higher erosion rate is expected because the soil is completely exposed to the impact of irrigation drops and the particle transport by runoff. CMT recorded an average erosion rate of 24.70 t ha^-1 year^-1. The VCT recorded the lowest average erosion values (18.97 t ha^-1 year^-1). However, there were no significant differences between these two. These results show that there is a significant decrease in soil erosion with the presence of vegetation cover. However, there was no effect due to the cover management (e.g., pruning). These results are consistent with those found by Bai and Cui (2021) in China, where bare soil had higher erosion rates compared to soils covered with corn and soybean crops, the latter presenting lower erosion values. These results show the importance of the barrier effect provided by soil vegetation covers and are in accordance with the coverage factor of the greenhouse experiment. In Figure 4, the BST presents a value of one (1) in the C factor, indicating that the soil is completely exposed to erosive processes. For the CMT, the C factor ranged between 0.88 and 1, showing a low level of soil protection, while the VCT presents the lowest values of C factor (0.55-0.63), indicating a greater potential of protection against erosion, although no significant differences between the latter two treatments were observed.

Figure 3.: Distribution of soil erosion under greenhouse conditions. BST: Bare soil; CMT: Conventional management; VCT: vegetation cover.

Figure 4.: Protective effect of soil covers. BST: Bare soil; CMT: conventional management; VCT: vegetation cover.

Field evaluations

The accumulated weekly precipitation is shown for the evaluation period (weeks 38 to 50 of 2022). This period coincided with the rainy season that occurs in the region at the end of the year. Consequently, precipitation events were observed every week during the field experiment (Fig. 5).

Sedimentation boxes. During the field evaluations, statistical differences were found among the treatments (P=0.02126). The treatments exhibited the following order of erosion rates: BST > CMT > VCT. The mean erosion rate of BST was classified as high (50-80 t ha^-1 year^-1), while those measured in VCT and CMT did not present significant differences and were classified as medium rates (25-50 t ha^-1 year^-1) based on the classification by Han et al. (2021).

The field evaluations showed the same behavior seen in the greenhouse experiment (Stage I); there was a significant decrease in soil erosion rates with vegetation cover, but no statistical difference was found in the type of management given to the covers. These findings are in accordance with those of Chen et al. (2019) in China, who reported an exponential decrease in erosive rates with increasing cover crop percentages, explaining the lower average values of VCT during the evaluations. Without active management, cover crops exhibit higher leaf density which provides greater barrier effects. Huerta-Olague et al. (2018) also mention that the covers generally tend to decrease the runoff and erosion rates. In Mexico with four cover crops, these authors found an exponential decrease in soil erosion with all of them. Additionally, they mention that the plant structure plays a crucial role in soil protection, with denser crops demonstrating greater abilities to protect the soil against erosive agents. The effect of covers in reducing erosion rates has been reported by various authors (Beniaich et al., 2023; Hou et al., 2020; Koirala et al., 2019; Xu et al., 2019; Zhao et al., 2020). Although there are no significant differences between VCT and CMT, channels with VCT management may present negative issues in operation, as these can obstruct water flow and decrease the speed at which water moves through the channel (Bond et al., 2020; Salazar, 2010; Tang et al., 2023; Zhao et al., 2019). This situation can lead to adverse effects, especially during periods of heavy rainfall when the channels must function properly to prevent flooding and damage to plantations. Therefore, proper maintenance is recommended, including pruning and weed removal, to reduce soil erosion and preserve channel functionality.

Figure 5.: Accumulated weekly precipitation during the weeks 38 to 50 of 2022 (September-December) corresponding to the field evaluation period.

Table 1.: Soil texture and organic matter content in the evaluated channels.

OM (%): soil organic matter content; BST: Bare soil; CMT: conventional management; VCT: vegetation cover; C: clay; CL: clay loam.

Erosive potential with USLE

The soils in the study area had an A horizon between 0 and 7 cm; based on the classification of USDA (USDA (2018)), the soils were categorized as degradation class 2, which indicates a loss between 25 to 75% of the thickness of the upper horizon of 20 cm, suggesting erosive processes in the area. Additionally, the soils exhibited a structure of fine to moderate sub-angular blocky aggregates, with the characteristics of texture and organic matter presented in Table 1.

Statistical differences in potential erosion values were found between the treatments (P=0.0317, Tab. 2). Based on the classification presented by Han et al. (2021), all treatments fell within the very high erosion risk (80-150 t ha^-1 year^-1).

Table 2.: Erosive potential of the evaluated soil.

BST: Bare soil; CMT: conventional management; VCT: vegetation cover. Different letters denote significant differences according to the Tukey HSD (P≤0.05).

These results indicated that the CMT treatment had the highest erosion potential, followed by BST and, finally, VCT. This suggests that external conditions besides the presence of covers make the soil susceptible to erosion. A Spearman correlation revealed that the LS factor had a correlation value of 0.66 and K factor of 0.46, which indicates that these two factors cause higher susceptibility to soil erosion, regardless of the coverage management. Belayneh et al. (2019) mentioned that the erosion rates in the Gumara watershed have a high relationship with its slopes, where steeper slopes result in higher erosion values. Additionally, Han et al. (2021) found that higher erosion rates are attributed to heavy precipitation (R), erodibility (K), and soil slopes (LS). Koirala et al. (2019) also refer to the increase in erosion with increasing land slopes.

These findings highlight that any erosion control strategies should use a comprehensive approach, considering the implementation of vegetation covers and the soil and landscape factors that contribute to the erosive potential. Therefore, a successful conservation solution must be adapted to the specific conditions of the area, incorporating practices that enhance soil properties such as soil structure and infiltration rates.